Offshore Wind Farm Design: offshore wind climate

Click here to load reader

  • date post

  • Category


  • view

  • download


Embed Size (px)



Transcript of Offshore Wind Farm Design: offshore wind climate

  • 1. Offshore Wind farm Design module 4: Offshore Wind Climate Course Offshore wind farm design OE 5662 Module 4 Offshore Wind Climate Background document Gerard van Bussel & Wim Bierbooms Section Wind Energy Faculty Aerospace Engineering September 2004 1
  • 2. Offshore Wind farm Design module 4: Offshore Wind Climate General information on wind climate1.1 Structure of the atmospheric boundary layerWind is moving air caused by pressure differences in the atmosphere, initiated bytemperature differences. At large heights (~1000 m) the wind is not perpendicular to theisobars but parallel to it due to the Coriolis force (caused by the rotation of the earth;perpendicular to the wind velocity). This wind is denoted the geostrophic wind.G : pressure gradient force Ug : geostrophic windC : Coriolis force U : wind at the groundW : drag force (friction) Fig. 1At lower heights friction will lead to wind shear: a reduction of the wind magnitude (especiallyin the so-called surface layer, the first few 100 meters above ground level) and a change inthe wind direction (especially above the surface layer: the Ekman layer). Fig. 2: Ekman layer in lower part of atmosphereThere are two causes of the friction:a) Mechanical: the friction is a function of the terrain roughness, characterised by the roughness length z0.b) Thermal: the friction depends on the stratification of the atmosphere: an unstable atmosphere will lead to mixing of air layers (the warmed air above the hot surface will rise and lead to mixing of air layers with different velocities). 2
  • 3. Offshore Wind farm Design module 4: Offshore Wind Climate1.2 Stability of the boundary layerThe picture below shows the general built up of the temperature and the pressure of the earthatmosphere. Fig 3: Temperature and pressure distribution in the atmosphere of the earthAn important aspect regarding the stability of the atmosphere is the temperature gradient inthe surface layer of the earth atmosphere. As can be seen there is a fairly constanttemperature gradient in the troposphere, the lower 10 km of the global atmosphere. A moredetailed evaluation of the pressure and temperature gradients in the troposphere is necessaryfor the determination of the stability of the atmospheric boundary layer. The stability of theboundary layer is the determining factor for the wind speed gradients that are experienced thefirst few 100 meters above ground level.1.2.1 Temperature and pressure in the boundary layersOne of the processes that are connected to vertical motions in the atmosphere is stability.This describes the tendency for the atmosphere to either resist or enhance vertical motions.The stability of the atmosphere is directly related to the changes of temperature with height,and evidently the changes of pressure with height.The changes of pressure with height in still air (no motions) are described by dp = ( z ) gdz (1)where is the density of the air, g the gravitational acceleration (9,806 m/s2) and z the verticalcoordinate.Now consider the temperature changes experienced by rising air. This is best done with theparadigm of a parcel of air. Such parcels of air are large enough to neglect the amount ofmixing with the surrounding air. As such a parcel of air rises, it moves into regions of lowerpressure. This means that the surrounding air is pushing on the parcel with less force. So theair in the parcel will expand, and the volume will become larger.When the air expands, the molecules must cover a larger volume. This means that the air inthe parcel must perform work to inhabit the increased volume. Since there is effectively notransfer of energy between the parcel of air and the surroundings, the work done by theparcel will result in lower temperature. Such a system is called adiabatic. The term simplymeans that no exchange of energy with the outside environment. The relation betweentemperature and pressure gradient is then given by: 1 c p dT = dp (2) 3
  • 4. Offshore Wind farm Design module 4: Offshore Wind Climatewhere cp is the specific heat at constant pressure (for dry air 1.005 kJ/(kg oK)) and T thetemperature [ oK]. The relation between temperature and height is thus: g dT = dz (3) cp An atmospheric boundary layer, in which the temperature distribution is according to theabove equation, is known as neutral. In such a neutral atmosphere a parcel of air will nothave the natural tendency to move vertically, since it experiences neutral buoyancy. Whensuch parcel rises its expansion and hence its temperature reduction matches exactly new theconditions (pressure and temperature) at the new height.1.2.2 Dry adiabatic lapse rateThe lapse rate, or vertical temperature gradient, expresses the change of the temperature ofthe atmosphere with altitude (height). The adiabatic lapse rate is the rate at which thetemperature of a parcel of descending or rising air would change due solely to compression orexpansion associated with elevation change.From the above equation (3) the dry adiabatic lapse rate can be determined. This is thechange of temperature a parcel of unsaturated air will experience in a neutral boundary layerwhen it is moved to another altitude:. dT g = = 9.76 [ o K/km] (4) dz cpThe dry adiabatic lapse rate, applicable for air containing no liquid water, is thus 9.76 K ofcooling per 1000 meters that the air rises. It is thus valid for parcels of air in which nocondensation of water occurs.1.2.3 Stable, neutral and unstable atmospheric boundary layersAn atmospheric boundary layer, in which the actual temperature distribution is according todry adiabatic lapse rate, is known as a neutral boundary layer. In practice the atmospheric (orenvironmental) lapse rate (the variation of temperature with height) will differ from the dryadiabatic lapse rate. If the gradient is less negative than the adiabatic value of 9.76 oK/km,then the atmospheric boundary layer is stable. Vertical movement of parcels of air arestabilised by the actual temperature gradient, since they cool adiabatically. Such situation iscalled a stable boundary layer.When the actual gradient is more negative that the dry adiabatic lapse the boundary layerbecomes unstable. A favourable ambient temperature encourages vertical movements of airparcels. A rising parcel will become relatively warmer with respect to its environment, andthus its buoyancy increases and it keeps rising. 4
  • 5. Offshore Wind farm Design module 4: Offshore Wind Climate Fig 4: Lapse rate in a stable and in an unstable atmospheric boundary layer.A stable atmosphere can usually be found above a cold surface, where hot surfaces willstimulate the development of unstable atmospheric boundary layers. Warmed air above sucha hot surface will rise and will lead to mixing of air layers with different velocities.1.2.4 Wind speed distribution in the atmospheric boundary layer. Fig. 5The (mean) wind speed distribution in a neutral atmospheric boundary layer can best bedescribed using a logarithmic profile. The following general equation is valid for the (mean)wind speed at some height V(h) with respect to the wind speed at a reference height (e.g. awind speed measured with a meteo mast at a certain height href ): ln( h / z 0 ) V ( h ) = V ( href ) (5) ln( h / z ) ref 0 The logarithmic profile is a result of a physical description of the boundary layer equations;hence it is preferred above a power law, which is sometimes used in e.g. standards.The value z0 is called the roughness height and is a measure for the surface roughness.Above land this roughness height typically varies between z0 = 0.03 m. and z0 = 1.00 m.At sea the roughness height is significantly less. A typical value at open sea is z0 = 0.0002 m.Note that the value of z0 is not a direct measure of the physical height of elements at thesurface, but is a fitting constant in order to fit the logarithmic profile to the observed profile athigher altitude.For non-neutral boundary layers a correction is usually applied upon the logarithmic windprofile: u* h V (h ) = ln z0 (6) 5
  • 6. Offshore Wind farm Design module 4: Offshore Wind Climate1.3 Estimation of the mean wind speed at a given locationIt is not possible to determine a map of the mean wind speeds due to the large localvariations of the roughness length. To overcome this problem a so-called potential windspeed has been defined: this is the wind speed at a standard height of 10 m which wouldexist in case the local roughness length would be equal to the standard value of 0.03 m. TheKNMI has processed such a map for the Netherlands based on several years of windmeasurements. Similar maps exist for other countries. Fig. 6As can be seen in the figure, there is a missing category, form 6.2 to 7 m/s. This is caused bythe way in which this potential wind speed map has been produced. Wind velocities at largerheights have been translated downwards to the reference height of 10 m with a roughnesslength of 0.03 m at land, but with a roughness length of 0.0002 m at sea. This causes thediscrete step between 6.2 and 7 m/s for coastal locations.1.4 Wind statisticsWind is extremely variable: it fluctuates continuously in speed and direction. This variabilityextends over an enormous range of time scales. This variability results from the manydifferent physical phenomena which influence the wind; each has its own characteristiceffects and occurs over a characteristic range of time (and length) scales. The figure shows a(fictional, similar to Van der Hoven) power spectrum of wind speed variations (showing thecontribution of different time scales to the total variability). Variability of the wind in time daily patterns (often thermal driven): contribution to wind speed sea breeze variability mountain slope winds turbulence, related to terrain, obstacles storm fronts (gusts) large scale weather systems high/low pressure systems seasonal variations spectral gap time scal year month day hour minute second 100-1000 km < 1 km space Fig. 7 6
  • 7. Offshore Wind farm Design module 4: Offshore Wind Climate On the scale second to minutes : stationary turbulence, caused by friction of the wind with the surface, by wakes of obstacles, by thermal instability (warm air rises); non stationary turbulence related to cold fronts, hail storms etc. On the scale of hours to days: passing of the above mentioned non stationary phenomena; thermally driven daily circulations as sea breezes, mountain slope winds On the scale of days to weeks: changes due to development and passing of large scale weather systems On the scale of weeks to months: seasonal differences, related to regional and global temperature differences; On the scale of years to years: year to year differences caused by cyclic solar activity, El Nino type of phenomena, long term trends eg. global warming !).The picture shows that time scales between many minutes and few hours contribute muchless to the overall variability than smaller and especially larger time scales. This means that ifone compares an one hour average with the next one hour average, these values are likely tobe more similar than comparing two successive week averages or second averages. The lowvariability around these middle time scales is called the spectral gap. The very existence ofthe spectral gap makes it possible to split the analysis of wind into two main parts A and B,where A is the variability due to turbulence and where B is the variability due toclimatology. It has turned out to be convenient to separate these two parts by using a 10minute average (or up to 1 hour, this choice is not very critical).The frequency distribution of the mean wind speed exhibits an important characteristic: it canbe assumed to be distributed according to the cumulative Weibull distribution function whichis determined by two parameters only: k F (U ) = 1 e (U / a ) (7) V time series of 10-min. means time number pdf Weibull V V histogram probability density function Fig. 8Here F is the probability of a wind speed smaller than U, a is the Weibull scale factor and k isthe Weibull form or shape factor. Another, maybe more familiar form is the following: k k U k f (U ) = e (U / a ) (8) UaHere f is the continuous probability density function of wind speed U. The Weibull parametersa and k can be calculated by a fit matching the observed data (measured or hindcast). Thiscan be done for each of the twelve wind direction sectors (of 30 degrees each).The parameter a is proportional to the mean wind speed with a given k. 7
  • 8. Offshore Wind farm Design module 4: Offshore Wind ClimateBy straightforward combination of the Weibull distribution and the power curve of the windturbine (power as function of the wind speed at hub height; both averaged over 10-min) theannual energy yield can be determined. Power kW contribution to annual energy production kWh V hours/yea r x = V total area = annual energy production V Fig. 9There are basically two ways to obtain the Weibull distribution for a specific site: derive fromexisting data or measure. To derive from existing data is of course much cheaper and quickerthan performing measurements. Because of the spatial variability, one can not assume thatthe wind speed at location A is equal to that of location B, even when they are only kilometresapart (effect of terrain shape, roughness, and obstacles). Correction methods have beendeveloped to take account of these differences. These methods (grouped here under thename of wind atlas correction methods) have their limitations and are increasinglytroublesome with increasing complexity of the terrain. Making measurements is than anoption, in particular when wind atlas methods are expected to be not reliable enough and/orwhen the wind farm investment is so large that highly accurate data are needed. problem: solution: Windatlas + correction use existing data data not use of correction methods representative methods (fetch, WAsP obstacles on land) Other two basic sources options not possible to correlate with other measure long sites which have Measure perform local enough to obtain reliable long term Correlate measurements good statistics statistics Predict MCP Fig. 10Because the wind is so variable, also from year to year (annual average wind speed from oneyear can easily differ 20 % from another year) one needs to measure for many years to obtaina reliable average. This is both time consuming and costly. As a solution, the so-called MCP(Measure Correlate Predict) methods have been developed, which correlate themeasurements with simultaneous measurements at a site where long term statistics areavailable. The required measurement period can thus be reduced. MCP methods have alsotheir limitations, the obvious one being that the method requires a correlation (physicalrelation) between both sites. Without (or with only little) correlation it just spoils the quality ofthe short term measurements. When this is the case, one just has to measure long enough todevelop a reliable long term picture. One should realise that both methods, in one way oranother, need to use existing long term reliable wind statistics: MCP to transform the short- 8
  • 9. Offshore Wind farm Design module 4: Offshore Wind Climateterm (even one year is short term) measurements to long term and wind atlas methods to givethe basis input. It is of course also useful to combine both methods and use the short termcorrelation to validate/calibrate theoretical methods.In the offshore zone, close to the coast, several factors complicate the assessment of windprofile and wind resource. It can be expected that stability effects are important; very lowroughness and large differences in temperature structures, both over the day and over theseasons. Three different cases are for instance: In the summer sea breezes can developduring the day and with onshore wind the air is in equilibrium with water surface temperature(1). The return flows at night results in relatively warm air flowing out over colder water (2). Inthe winter the water is usually warmer than the land, so when the wind flows from land to sea,cold air flows over warm water (3). It has turned out that current wind atlas methods (~1999/2000) can not accurately predict the wind resource in the coastal zone. Only when oneis far enough from the land, the methods can more or less be applied, provided account istaken for the fact that the surface roughness is to some extent depending on the wind speeditself (high waves at high wind speeds give a somewhat higher roughness). Developmentwork is yet going on to improve the modelling in the coastal zone, so-called CoastalDiscontinuity Models. 9
  • 10. Offshore Wind farm Design module 4: Offshore Wind Climate1.5 TurbulenceVariation of the wind speed within a period of 10-min is called turbulence. These fluctuationshave a Gaussian (normal: bell-shaped) distribution. The ratio between the standard deviationand the 10-min. mean wind speed is called turbulence intensity. Turbulence does not effectthe energy yield but it is vital for the dynamic wind turbine loading. Fig. 11The characteristic time scales of turbulence varies between a second to minutes; thecorresponding length scales range from a meter to hundreds of meters. The spatial variationof turbulence is felt by a rotating wind turbine blade as temporal ones; this phenomenon isdenoted rotational sampling. It results in a dynamic wind loading with main frequencies equalto the rotor frequency (so-called 1P) and its multiples (2P, 3P, nP). Apart from turbulencethe following wind loads result in periodic wind loads which may lead to fatigue: wind shear,yawed flow and tower shadow. Note: also gravity acts as a periodic load (in the so-calledlead-lag direction: parallel to the rotor plane). Rotational sampling of turbulence eddies rotor tower shadow Fig. 12Nowadays, it is common practice to generate 3D stochastic wind fields, for several mean windspeeds, to account for turbulence in the simulation of wind turbines. For such a simulation,the turbulence spectrum should be known as well as the turbulence intensity. The number ofoccurrences of each mean wind speed during the design lifetime of the wind turbine followsfrom the Weibull distribution.1.6 Extreme wind speeds 10
  • 11. Offshore Wind farm Design module 4: Offshore Wind ClimateThe KNMI has also compiled a map of extreme wind speeds (10-min. or 1 hour mean values).Usually a return period of 50 years is applied; i.e. the extreme wind speeds occur on averageonce in 50 years. Fig. 13For the determination of the extreme wind loads, the turbulence on top of the extreme windspeeds must be taken into account. For this purpose, standards specify deterministic windgust shapes. Ongoing research aims at a more precise, stochastic description of gusts. Overview of existing offshore wind information1.7 Overview of existing dataFor information on the offshore wind climate several sources of data are at hand.However, the combination of detailed and long-term data of sufficient quality and availability isvery rare. Therefore different types of data will probably have to be combined to achieve goodresults.There are three main criteria that are important. For each type of data source the extent withwhich is complied to these criteria is shown in Table 1.. Accuracy Duration ResolutionPressure data -- ++ -Data VOS - + ++Light-ship data +/- ++ -Satellite data +/- - +Platform data ++ +/- +NESS ++ + ++ Table 1: Overview of characteristics of several data sources1.7.1 Pressure dataThe main problem with using pressure data is the fact that a conversion step is necessary toobtain wind speed data. Theoretically the geostrophic wind (also called free wind, because itis the wind that does not depend directly on the characteristics of the underlying surface) canbe derived simply from pressure gradients. For the North Sea this has been done byBrresen, leading to the Wind Atlas for the North Sea and the Norwegian Sea. The underlyingpressure data set had a duration of 27 years (1955-1981). The grid size was 150 km by 150km, and the temporal resolution was 6 hours. The geostrophic wind speeds were reducedfurther to 10-minute mean wind speeds at 10 m height by using a geostrophic drag law. 11
  • 12. Offshore Wind farm Design module 4: Offshore Wind ClimateThe JOULE-project POWER uses similar methods to obtain geostrophic wind speeds over allEuropean seas. The reduction to near-surface winds is carried out using WAsP.In the table the accuracy does not score well, because the translation from pressureobservation to near-surface wind speed inherently gives rise to uncertain results. The twomain advantages of the use of pressure data are: the duration (can be over 100 years) andthe large coverage.1.7.2 Data Voluntary Observing ShipsThe data of Voluntary Observing Ships are reliable and have been checked since 1960. Theycover the entire globe and have high spatial resolution (0.1 0.1). The individualobservations probably have flaws, due to the variety of observers and methods of observation(visually or using instruments). It is not possible to check every observation, although thereare standard procedures. But due to the abundance of the data they may prove very valuable.The data consist of wind speed and direction, wave parameters, visibility, temperatures of airand sea water and atmospheric pressure.Several studies have been conducted using VOS data as a starting point. One is a JOULEproject by Germanischer Lloyd and Garrad Hassan leading to the report Study of OffshoreWind Energy in the EC. For this project the Deutsche Seewetteramt has (re)analysed manyVOS-data and derived a crude wind map for all European seas with wind speeds re-scaled to25 m height.In the past KNMI has published a report by Korevaar using VOS showing several results ofanalyses. E.g., exceedance of wind speeds expressed in Beaufort and of wave heights in mfor the North Sea in several seasons and for a whole year. Furthermore, extremes of the windand waves are given in return periods of 50 years. KNMI is willing to make the data availableand suitable for an update of these analyses. The data comprises 40 years: 1960-2000.1.7.3 Light-ship dataData from light-ships have been used quite extensively in the past to gather information aboutthe offshore wind and wave climate. Another report by Korevaar was devoted to this topic.This type of data was also used for Windklimaat van Nederland.Light-ships are stationary and usually are installed for several decades. In recent years theyhave been removed and were replaced by unmanned light isles or otherwise. Theobservations at light ships were made by visual inspection, with the use of the Beaufort scaleaccording to its strict definition. In a later stage the conversion from Beaufort classes to near-surface wind speeds was investigated and established.Another study was conducted by RWS, in which observations of the period 19071980 wereanalysed from three light-ships: Haaks & Texel, Schouwenbank & Goeree and Noordhinder.Finally, the two light-ships Noord-Hinder and Texel were also described.1.7.4 Satellite dataThe technique of remote sensing has developed very rapidly in recent years. The use ofsatellites to gather data is now common practice. Over water surfaces it is possible tomeasure wind speeds by interpreting the wave appearance. This technique has theadvantage of great spatial coverage, but the disadvantage of poor accuracy and temporalcharacter (snapshots). Furthermore, satellite images have become available during recentyears so the length is not very long. However, this type of data source will become more andmore widely used and may provide additional information to other available data sources atthis moment.1.7.5 Data from platforms and other offshore sitesSince the beginning of the eighties, just when the light-ships were being removed, KNMI andRWS have started to install meteorological instruments on various platforms and otheroffshore constructions called the Measuring Network North Sea. This Network is unique andsupplies invaluable information. Both wind and wave characteristics are measured, and nowalready during more than 10 years at several locations. RWS is mainly responsible foroperation and maintenance, KNMI for the calibration of the instruments, the quality of the dataand the storage for later use.About 10 years ago the first research project was undertaken to make use of the data of aplatform for wind energy applications. 12
  • 13. Offshore Wind farm Design module 4: Offshore Wind ClimateIn about the same period a similar network was established in the province of Zeeland byRWS. It is called the Monitoring Network ZEGE (ZEeuwse GEtijdewateren). In stead of usingexisting structures custom made measuring posts were used. Currently about 10 are in usethat are equipped with anemometers. Also wave measurements are carried out at severallocations by buoys. The data are gathered and stored for later use. Although the main focus ison the inland waters of Zeeland, some posts are located in the North Sea. Data form ZEGEwere used for the feasibility study for the 100 MW Near Shore demonstration project.In England the oil platform West Sole has been equipped for some time with instruments atseveral heights up to about 80 m above Mean Sea Level (MSL). The feature that makes thisconfiguration unique is the fact that turbulence intensities were measured as well as windspeeds. Several publications have been devoted to the results of the measurement campaignconducted at West Sole. However, one conclusion drawn from a comparison with otherplatform data was that the situation just east of the English coast is not the same as west ofthe Dutch coast in terms of temperatures and therefore stability. In prevailing westerly windsthe fetch at West Sole is mainly over land, while near the Dutch coast it is mainly over sea. Fig. 14 ID Name Code Northing Easting Anemometer Period Water height1 depth NL20 IJmuiden-harbour IJM 522747" 043322" 18.5 m 19871997 n.a. NL24 IJmuiden depot IJM 523330" 040330" n.a. 19871997 21 m NL23 Measuring post Noordwijk MPN 521623" 041750" 27.6 m 19871997 18 m NL22 Light Isle Goeree LEG 515529" 034006" 38.3 m 199119972 22 m NL13 Euro platform EUR 515955" 031635" 29.1 m 19871997 32 m NL21 K13 platform K13 531304" 031313" 73.8 m 19871997 30 m NL01 Auk platform AUK 562359" 020356" 103.3 m 19871995 85 mNote: The location of the AUK platform is not shown in the figure as it is further north nearNorway.Table 2: Characteristics of the selected offshore locations with concurrent wind speed and wave observations of the Measuring Network North Sea.1.7.6 Availability of dataConsiderable effort has been devoted to obtain information about the availability of data.Knowing the existence of databases and of the owner or user is not enough to actually gethold of them. In general three routes for various databases can be distinguished. 13
  • 14. Offshore Wind farm Design module 4: Offshore Wind ClimateThe main institution collecting and distributing data in The Netherlands is KNMI. They haveestablished an online database which can be assessed for wind energy applications. Theproject in which this database is set up and can be used by external parties is called theHydra project ( Most of the informationconcerns onshore locations, but there is also data available for 7 offshore measuring posts.1.8 Overview existing knowledgeMany of the formulas given in this report depend on the stability of the boundary layer. Here itis assumed that the stability conditions are neutral. For the time being this seems justifiedbecause on average the influence of the stability on the wind profile seems to be very limitedaccording to measurements at on offshore location.By means of sensitivity studies, on energy yield and design loads, the validity of thisassumption can be investigated. Furthermore, neutral conditions correspond with high windspeeds, so the assumption will probably lead to conservative design loads.1.8.1 Wind resourceFor the economics of a wind farm, the estimate of the average energy production is essential.For an accurate estimate of the energy output the following information is important. First thestatistics of wind speeds are described, than an overview of methods to calculate the windspeed at hub height (wind profile) is givenStatistical descriptionOne of the main drivers for offshore siting of wind turbines is the higher mean wind speed. Inthe figure the Weibull distribution onshore and offshore are compared. 0.14 mean wind speed 6.5 m/s 0.12 (onshore) mean wind speed 8 m/s 0.1 (offshore, Baltic) probability [-] mean wind speed 10.1 m/s 0.08 (remote offshore, North Sea) 0.06 0.04 0.02 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 wind speed at hub height [m/s] Fig. 13The annual mean wind speed can vary considerably in different years. A standard deviation ofannual mean wind speed of 11% is found for offshore locations, which is large compared tothe 5.5% for onshore sites.In order to determine the long term mean wind speed a database of 30 years or more isnecessary; such long databases are not (yet) available for offshore locations. In case such adatabase becomes available for some locations, this will be also useful for locations nearby(for which only a limited data set is available). A correction factor can be determined based onthe ratio of the annual mean wind speed for the two locations. Such method is standardpractice for onshore situations. 14
  • 15. Offshore Wind farm Design module 4: Offshore Wind Climate Annual variation of the mean wind speed at six locations (see table 2) 10.5 10.0 9.5 10-m wind speed [m/s] 9.0 8.5 8.0 7.5 7.0 6.5 6.0 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 AUK EUR K13 LEG MPN IJM Fig. 14Wind profileUsually the wind speed at hub height has to be calculated from wind speed observations ormodels where the wind speed is given at a standard height of 10 m. The simplest relationshipto describe the wind profile (i.e. the horizontal wind speed U as function of height Z) is apower law: U ( Z1 ) Z1 = U (Z 2 ) Z 2 (9)This is a non-physical expression which gives reasonable results for the correct value of (usually in the order of 0.1, for offshore locations the exponent is equal to 0.11 according toGL). The IEC uses a value of 0.2, KNMI uses a value of 0.13 for standard height correctionsof offshore observations.A first step towards a physical description of the wind profile is the logarithmic law. Itincorporates the roughness length z0, and gives therefore more information, in this case aboutthe underlying surface: Z U ( Z1 ) ln u* U ( Z1 ) = ln 1 = ( ) Z1 z0 z0 U ( Z 2 ) ln ( ) Z2 z0 (10)The value of the Von Krmn constant is 0.4, while u* is the friction velocity, a measure forthe effectiveness of vertical exchange of momentum. In this expression it is assumed thatthere is no heat exchange (adiabatic conditions), and the wind profile is completelydetermined by mechanical turbulence. This assumption is certainly true for high wind speeds,and appears to be true on average for offshore conditions.Onshore the value of z0 depends on the terrain conditions only and is not influenced by thewind speed. It can of course be dependent of the wind direction because it reflects thecharacteristics of the upwind terrain only.In offshore conditions the value of z0 does depend on the wind speed. Due to the generationof waves by the wind there is a strong correlation between the wind speed and the roughnesslength. In a simple form this relation was formulated by Charnock as follows: 15
  • 16. Offshore Wind farm Design module 4: Offshore Wind Climate u* 2 z0 = g (11)Here g is the gravitational constant, and is a constant with a value of 0.01-0.02. In neutralconditions there are two equations, (5) and (6), with two unknowns (U and z0) that can besolved numerically. onshore offshore z0 depends on wave height H 120 (m) verification 100 needed for wind shear 80 60 40 20 0 5 5.5 6 6.5 7 7.5 8 8.5 9 V (m/s) Fig. 15Stability effectsFinally thermal effects can be introduced into the picture. In general this is done by extendingthe earlier given logarithmic profile relationship with an extra term: u* h1 h1 U ( h1 ) = ln z0 L (12)The term is the so-called stability function. It depends on the ratio of the height and theObukhov length L. The value of L gives an indication of the stability conditions. For negativevalues of L the atmospheric conditions are unstable (warm surface, cold air, much verticalmixing), for positive values of L conditions are stable (cold surface, warm air, little verticalmixing, strong stratification). If the absolute value is large (>1000) the conditions are (near-)neutral (adiabatic conditions).Although this framework appears to be rather simple (i.e. we now have three equations withthree unknowns which have to be solved numerically), there are a few snags.The first is that the form of the function m has to be determined, and the value of L has to becalculated. Basically the same equations are used for land or sea surfaces but with differentparametrisations and input parameters. In the literature several formulations for this can befound. For conditions over sea the following input parameters are needed: wind speed anddirection, air and seawater temperature. The second is that the validity of the usedexpressions is limited to a certain layer. In this layer (the atmospheric boundary layer) theconditions are subject to the physics as expressed in the previously given equations.However, the depth or height of this layer is dependent on several conditions. Especially inthe case of very stable conditions (small positive values of L) the atmospheric boundary layermay be very thin, less than the (upper) tip height of a large wind turbine. In this case strongwind shear may occur due to uncoupling of the wind speed close to the surface and abovethe height of the boundary layer. Therefore care has to be taken with the use of the givenequations and the interpretation of the results.Moreover, a model is always a simplification of the reality and is in itself a possible source oferrors. When it comes to exploring the outer limits of the model validity, it is important to have 16
  • 17. Offshore Wind farm Design module 4: Offshore Wind Climateexperiments to corroborate the theory. In the case of the theoretical description of a windprofile beyond 100 m above ground level there are hardly or no experimental data to rely on.Comparison of conditions onshore vs. offshoreLooking at the general situation onshore compared to that offshore the followingcharacteristics can be identified (see Table 3).The reason for differences is as follows: first of all the sea water has a higher heat capacitythan the soil onshore. Furthermore it is vital that the irradiated energy is mixed down. Thisresults in a sea surface temperature which is constant during the day, but has a seasonalpattern which lags behind the pattern of the air temperature over land. This will not influencethe energy yield as such but may be important with respect to the economic value of offshorewind energy on a free electricity market.On land the irradiated heat leads to the warming of only the top of the surface. During thenight this heat is radiated with long wavelengths, resulting in a cooling of the surface. Sotherefore the diurnal pattern of the surface temperature follows the solar irradiation (with atime lag of maybe a few hours). During the day the surface heats up leading to more mixingand therefore a wind speed maximum in the afternoon. onshore offshorediurnal pattern daily maximum uniformseasonal pattern less pronounced more pronouncedstability diurnal pattern seasonal patternwind profile unstable on average neutral on averagemean wind speed decreasing inland higher than on landTable 3: Comparison of atmospheric characteristics onshore and offshore Monthly variation of the mean wind speed at six locations (see table 2) 12 11 10-m wind speed [m/s] 10 9 8 7 6 5 Jan Feb Mrt Apr Mei Jun Jul Aug Sep Okt Nov Dec AUK EUR K13 LEG MPN IJM Fig. 16 17
  • 18. Offshore Wind farm Design module 4: Offshore Wind Climate Hourly variation of the mean wind speed at six locations (see table 2) 9.5 9.0 10-m wind speed [m/s] 8.5 8.0 7.5 7.0 6.5 6.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hour AUK EUR K13 LEG MPN IJM Fig. 17Looking at a larger scale two types of climates can be distinguished: offshore vs onshore. It isclear that in the coastal zone there is a transition between the two types where mixed featureswill occur, mainly depending on the wind direction. Some work has been devoted todetermine to what distance such a coastal zone extends, and it is generally believed to besome tens of kilometres.WAsPThe model WAsP by Ris is commonly used for calculating energy productions in manycountries. Several parameters used in the underlying model are tuned to yield correct resultson average during average conditions. In principle WAsP can be used for offshore purposes,but the tuning of the parameters is not adapted for offshore applications yet. However, severalpublications have shown encouraging results in certain particular conditions.Models at KNMIAt KNMI an operational model is being tested which is capable of describing the wind fieldoffshore. This means that wind speeds offshore can be predicted from onshore data. Thesesituations represent only a limited number of weather situations, but these are the mostdifficult to model. Westerly winds have a sea fetch which is modelled far more simply. Anotherinteresting trend is the increasing use of so-called mesoscale models. At KNMI the meso-scale model Hirlam is available and is undergoing constant development.1.8.2 Design Parameters FatigueThe design process requires different input of the environmental conditions. A distinction hasto be made between the design against fatigue and the design against extremes. For thefatigue calculations the distribution of the mean wind speed is mandatory. Moreover, for thesimulation of the turbulence, the spectra of the wind turbulence have to be specified. 18
  • 19. Offshore Wind farm Design module 4: Offshore Wind ClimateIn the different design codes, the spectra specification differs. There is no specific guidelinefor the offshore turbulence spectrum and by lack of sufficient and reliable offshoremeasurements, it is generally assumed that it can be expressed with the conventionalspectra. One example is the von Krmn spectrum for the longitudinal component of the windspeed: ~ f S uu 4f = ~ uu 2 (1 + 70.8 f 2 ) 5 / 6 (13)S uu is the auto power spectral density of the wind speed uu is the standard deviation of the wind speedf is the frequency (Hz)~f is the dimensionless frequency~ x Lu ff = U xHere, the turbulence length scale Lu can be defined as 25z 0.35 x Lu = z 0 .0063 0 (14)The turbulence intensity is the governing measure for fatigue loads. According to the ESDUthe turbulence intensity can be formulated as F ( z) I t ( z) = ln( z / z 0 ) (15)with F ( z ) = 0.867 + 0.556 log100 (z) - 0.246(log10 (z)) (it can be assumed that F(z)1). 2This equation is also given in the Danish Standard (DS 472). This expression is to bepreferred above i.e. the equations for the turbulence intensities as given in the IEC code,which are applicable for whole regions (indicated by WTGS classes). The latter are based onturbulence measurements at a very broad range of sites and conditions (e.g. atmosphericstability) which therefore show a lot of scatter. The surface roughness length is a function ofthe mean wind speed according to Charnocks relation, thus the turbulence intensity is alsodependent of the mean wind speed. The GL offshore regulation prescribes a turbulenceintensity of 12%. Measurements show a range of 8 to 12% depending on the mean windspeed and stability conditions.For the design of an offshore wind farm at some specific site, it is recommended to use theturbulence intensity at that site, in case measurements (of say at least 1 year) are available;otherwise equation (10) may be applied. A certification based on the actual site conditionsseems justified in respect to the large number of wind turbines of an offshore wind farm;application of the turbulence intensities as given in the IEC code will lead to far moreconservative designs. re d u c e d w in d s pe e d tu r b in e w a k e d o w n w in d tu rb ine : lo w e r w in d sp e ed : le s s p o w e r h ig h e r tu rbu le nce a n d in c rea se d w in d s h ea r: m o re fa tig u e in c re as e d tu r b u le n c e 19
  • 20. Offshore Wind farm Design module 4: Offshore Wind Climate Fig. 18The efficiency of the wind farm and the loads are strongly influenced by the wake structure.The increase of turbulence due to the presence of the wakes produced by the upstreamturbines is considerable and can be approximated as 0.4CT I eff = + I0 2 st 2 (16)where,CT is the thrust coefficient of the closest wake generating turbinest is the non-dimensional turbine separation x/D (D is the diameter of the rotor)I0 is the ambient turbulence of the free flow.For the calculation of the fatigue load the logarithmic wind profile mentioned above should beapplied. Furthermore, the variation of the wind direction with height should be taken intoaccount. For this purpose the geostrophic drag law may be applied (in line with WAsP). It isstandard to use such an expression for scaling of the wind speed and direction to hub heightfor energy yield calculations; it is yet to be determined how to incorporate this in a fatigueanalysis.Sensitivity studies show that the turbulence intensity is by far the dominant factor in fatigueloads.For simulation of the wind velocities at different locations the knowledge over the crosscorrelation function is indispensable. The non-symmetrical cross correlation function results ina complex spectrum; however the quad-spectrum is usually neglected.As is the case for spectra, there is no specific coherence model specified for the offshorewind, nor enough and reliable measurements. Conventional coherence models can be foundin the IEC draft or ESDU.TurbulenceDue to the lower mechanical friction at the sea surface less turbulence is created above thesea and turbulence levels are therefore lower than over land. The turbulence level is given by z i 1/ 3 u (12 0.5 L ) I( z ) = = U ln(z / z 0 ) (z / L) (17)where zi denotes the height of the planetary boundary layer, L the Obukhov length and z0 theroughness length.Under the condition that the boundary layer is in equilibrium with the underlying surface theseparameters can be derived using local conditions such as wind speed, sea and airtemperature. For offshore wind conditions (wind coming from sea) this assumption is true ifthe fetch of water is sufficiently long. In nearshore conditions with wind flowing from land, thisis not longer the case. Erbrink describes measurements at Meetpost Noordwijk, in situationswith Easterly winds, where there is hardly found any correlation between the turbulence andthe local wind speed and stability parameters. A better correlation is found when the land-seatransition is modelled and land conditions are used to initialise the calculations.Offshore wind measurements suggest that the wind profile is influenced by upstream landconditions up to 30 km offshore. The adaptation of turbulence to the underlying sea conditionsis expected to take place at an even slower pace. This suggests that one should considerland-based turbulence in those cases where winds have their origin over land.Measurements of turbulence in offshore conditions are scarce. Larsen et al. have analysedwind data from a wind measuring mast at Vindeby wind farm. The mast is situated in shallowwater erected very close to the shoreline. The data are recorded at a level of 37.5 m. In thepaper only the sea fetch is considered. The sea fetch is characterised by having more than 15km of sea upstream. The available data constitutes 5566 10-minute time series with anoverall mean wind speed equal to 7.92 m/s. A design turbulence level (weighted averagebased on fatigue loads) is derived from the measurements. It should be noted that these 20
  • 21. Offshore Wind farm Design module 4: Offshore Wind Climatevalues have been derived for shallow water and a limited sea fetch of 15 km, and that theresults are not automatically valid for Dutch offshore conditions. Fig. 19: Design standard deviation as a function of wind speed found at Vindeby (offshore sectors, sea fetch >15 km).Low Level JetsA Low Level Jet (LLJ) is a phenomenon in which a wind maximum occurs at relatively lowlevels in the atmosphere (150 m).Onshore, the phenomenon of the Low Level Jet typically occurs during stable night-timeconditions, in which case it is also known as a nocturnal jet. During clear nights, a stable, coldlayer develops near the ground. In this layer, turbulence is damped and the vertical exchangeof momentum is limited. The layer above the stable layer is effectively de-coupled and not anylonger pulled back by the lower layer. It therefore accelerates, sometimes reaching velocitiesthat are even higher than the geostrophic wind speed. The maximum wind speed occurs atheights between 100 and 150 m. The nocturnal jet is hence a time-dependent phenomenon inwhich the development of a stable boundary layer is a prerequisite. Since the diurnal courseof the temperature over the sea is much smaller than over land the likelihood of such a typeoffshore nocturnal jet seems small. Fig. 20An offshore LLJ is reported in the Baltic Sea, where it is rather common. The figure showstwo wind speed profiles taken from a pibal tracking. The solid line shows the LLJ. The figureshows that during the phenomenon the wind speed can reach a considerable value. Oneconsequence of the non-logarithmic behaviour is that extrapolations from one height to theother height result in erroneous wind speeds and that the wind potential in the Baltic isunderestimated. Another consequence is the rather strong wind shear, which can an effect onthe wind turbines life. 21
  • 22. Offshore Wind farm Design module 4: Offshore Wind Climate1.8.3 Maintenance and installationThe main concern for maintenance and installation operation is the accessibility delimited bythe Met-Ocean variables. Generally for a location with 10 to 20 km distance from the coast,an accessibility of 75% percent can be assumed.For development of an optimal maintenance strategy, detailed data has to be available inorder to determine the weather windows. Depending on the means of transport and theequipment employed the specific criteria for these windows are different; maxima allowedvalues of the wind speed and wave height (and visibility). LiteratureMain parts of this document are extracted from:J.P. Coelingh (ed.), Wind and wave data compiled for the DOWEC concepts study, SectionWind Energy IW-00162R, Delft University of Technology, 2000.Recommended literature on wind climate are (focus on onshore):Wieringa, J. , Rijkoort, P.J., Windklimaat van Nederland [In Dutch], KNMI, 1983.Troen, I., Petersen, E.L., European Wind Atlas, Ris National Laboratory, 1989.Petersen, E.L. et al., Wind Power Meteorology, Wind Energy, number 1 and 2, 1998.Information on the offshore wind climate can be found in the OWEMES proceedings, finalreports of general offshore wind energy projects and specialised papers, e.g.:Coelingh, J.P., Van Wijk, A.J.M., Holtslag, A.A.M., Analysis of wind speed observations overthe North Sea, J. Wind Engin. Industr. Aerodyn., Vol. 61, Nr. 1, pp. 5169In the near future information will become available on several European offshore windenergy projects: POWER, MAST-III and ENDOW (wake effects offshore).Wind and offshore wind turbine load measurements are currently performed at one of the twooffshore wind turbines at Blyth (UK); at a.o. Vindeby and Horns Rev (Denmark) windmeasurements have been performed. Similar measurements are scheduled for the comingDutch Nearshore Windfarm (for the coast of Egmond).For an introduction to the aerodynamic loading of wind turbines see a text book on windenergy, e.g.:Freris, L.L. (ed.), Wind Energy Conversion Systems, Prentice Hall, 1990.Figures1, 2, 5,6 and 13 from:Wieringa, J. , Rijkoort, P.J., Windklimaat van Nederland [In Dutch], KNMI, 1983.4 "The Stratosphere." Met Office of the United Kingdom., 9 and 18: Renewable Energy Course Materials Tackling Market Barriers, Altener ProjectAL 98-022.10: Fatigue Design Seminar, UCL/nCode, UK.Robert Gasch (ed.), Windkraftanlagen,Stuttgart 199311: Department Of Energy, USA.13 and 15: Martin Khn, Dynamics and Design Optimisation of Offshore Wind EnergyConversion Systems, PhD report 2001.14, 16 and 17: J.P. Coelingh (ed.), Wind and wave data compiled for the DOWEC conceptsstudy, Section Wind Energy IW-00162R, Delft University of Technology, 2000.. 22